The structure of the aquaporin-1 water channel: a comparison between cryo-electron microscopy and X-ray crystallography

Three different medium-resolution structures of the human water channel aquaporin-1 (AQP1) have been solved by cryo-electron microscopy (cryo-EM) during the last two years. Recently, the structure of the strongly related bovine AQP1 was solved by X-ray crystallography at higher resolution, allowing a validation of the original medium-resolution structures, and providing a good indication for the strengths and limitations of state of the art cryo-EM methods. We present a detailed comparison between the different models, which shows that overall, the structures are highly similar, deviating less than 2.5 A from each other in the helical backbone regions. The two original cryo-EM structures, however, also show a number of significant deviations from the X-ray structure, both in the backbone positions of the transmembrane helices and in the location of the amino acid side-chains facing the pore. In contrast, the third cryo-EM structure that included information from the X-ray structure of the homologous bacterial glycerol facilitator GlpF and that was subsequently refined against cryo-EM AQP1 data, shows a root mean square deviation of 0.9A from the X-ray structure in the helical backbone regions.     

Functional interactions at the interface between voltage-sensing and pore domains in the Shaker K(v) channel

Voltage-activated potassium (K(v)) channels contain a central pore domain that is partially surrounded by four voltage-sensing domains. Recent X-ray structures suggest that the two domains lack extensive protein-protein contacts within presumed transmembrane regions, but whether this is the case for functional channels embedded in lipid membranes remains to be tested. We investigated domain interactions in the Shaker K(v) channel by systematically mutating the pore domain and assessing tolerance by examining channel maturation, S4 gating charge movement, and channel opening. When mapped onto the X-ray structure of the K(v)1.2 channel the large number of permissive mutations support the notion of relatively independent domains, consistent with crystallographic studies. Inspection of the maps also identifies portions of the interface where residues are sensitive to mutation, an external cluster where mutations hinder voltage sensor activation, and an internal cluster where domain interactions between S4 and S5 helices from adjacent subunits appear crucial for the concerted opening transition.     

Structural and energetic analysis of activation by a cyclic nucleotide binding domain

MlotiK1 is a prokaryotic homolog of cyclic-nucleotide-dependent ion channels that contains an intracellular C-terminal cyclic nucleotide binding (CNB) domain. X-ray structures of the CNB domain have been solved in the absence of ligand and bound to cAMP. Both the full-length channel and CNB domain fragment are easily expressed and purified, making MlotiK1 a useful model system for dissecting activation by ligand binding. We have used X-ray crystallography to determine three new MlotiK1 CNB domain structures: a second apo configuration, a cGMP-bound structure, and a second cAMP-bound structure. In combination, the five MlotiK1 CNB domain structures provide a unique opportunity for analyzing, within a single protein, the structural differences between the apo state and the bound state, and the structural variability within each state. With this analysis as a guide, we have probed the nucleotide selectivity and importance of specific residue side chains in ligand binding and channel activation. These data help to identify ligand-protein interactions that are important for ligand dependence in MlotiK1 and, more globally, in the class of nucleotide-dependent proteins.     

Molecular regions underlying the activation of low- and high-voltage activating calcium channels

We have studied two aspects of calcium channel activation. First, we investigated the molecular regions that are important in determining differences in activation between low- and high-voltage activated channels. For this, we made chimeras between the low-voltage activating Ca_V3.1 channel and the high-voltage activating Ca_V1.2 channel. Chimeras were expressed in oocytes, and calcium channel currents recorded by voltage clamp. For domain I, we found that the molecular region that is important in determining the voltage dependence of activation comprises the pore regions S5-P as well as P-S6, but surprisingly not the voltage sensor S1–S4 region, which might have been expected to play a major part. By contrast, the smaller, but still significant, modulating effects of domain II on activation properties were due to effects involving both S1–S4 and S5–S6 but not the I/II linker. Second, during channel activation we studied movement of the S4 segment in domain I of one of the chimeras, using cysteine-scanning mutagenesis. The reagent parachloromercuribenzensulfonate inhibited currents for mutants V263, A265, L266 and A268, but not for F269 and V271, and voltage dependence of inhibition for residue V263 indicated S4 movement, which occurred before channel opening. The data indicate movement outwards upon depolarisation so as to expose amino acids up to residue 268 in S4.Junying Li and Louisa Stevens contributed equally to this work.     

Sodium channel functioning based on an octagonal structure model

The complete amino acid sequence of a sodium channel from squid Loligo bleekeri has been deduced by cloning and sequence analysis of the complementary DNA. A unique feature of the squid sodium channel is the 1,522 residue sequence, approximately three-fourths of those of the rat sodium channels I, II and III. On the basis of the sequence, and in comparison with those of vertebrate sodium channels, we have proposed a tertiary structure model of the sodium channel where the transmembrane segments are octagonally aligned and the four linkers of S5–6 between segments S5 and S6 play a crucial role in the activation gate, voltage sensor and ion selective pore, which can slide, depending on membrane potentials, along inner walls consisting of alternating segments S2 and S4. The proposed octagonal structure model is contrasted with that of Noda et al. (Nature 320; 188–192, 1986). The octagonal structure model can explain the gating of activation and inactivation, and ion selectivity, as well as the action mechanism of both tetrodotoxin (TTX) and -scorpion toxin (ScTX), and can be applied not only to the sodium channel, but also to the calcium channel, potassium channel and cGMP-gated channel.The authors would like to express our cordial acknowledgments to Dr. Hideo Tani (Kowa) and Drs. Masahiko Fujino and Haruo Onda (Takeda Pharmaceutical) for their kind support for us to utilize their experimental facilities for DNA cloning and as well as for their stimulating and helpful discussions. We also thank Drs. Toshio Iijima, Michinori Ichikawa, Kiyonori Hirota, Messrs. Tadashi Kimura and Osamu Shono and all our colleagues (Supermolecular Science Division, Electrotechnical Laboratory) for their kind support to collect and isolate optic lobes from live squid. We greatly thank Professors Takuji Takeuchi (University of Tohoku) and David Landowne (University of Miami) for their illuminating discussions and valuable comments.     

Graded activation of CRAC channel by binding of different numbers of STIM1 to Orai1 subunits

The Ca²⁺ release-activated Ca²⁺ (CRAC) channel pore is formed by Orai1 and gated by STIM1 after intracellular Ca²⁺ store depletion. To resolve how many STIM1 molecules are required to open a CRAC channel, we fused different numbers of Orai1 subunits with functional two-tandem cytoplasmic domains of STIM1 (residues 336-485, designated as S domain). Whole-cell patch clamp recordings of these chimeric molecules revealed that CRAC current reached maximum at a stoichiometry of four Orai1 and eight S domains. Further experiments indicate that two-tandem S domains specifically interact with the C-terminus of one Orai1 subunit, and CRAC current can be gradually increased as more Orai1 subunits can interact with S domains or STIM1 proteins. Our data suggest that maximal opening of one CRAC channel requires eight STIM1 molecules, and support a model that the CRAC channel activation is not in an “all-or-none” fashion but undergoes a graded process via binding of different numbers of STIM1.     

Identification of Amino Acid Residues in the α, β, and γ Subunits of the Epithelial Sodium Channel (ENaC) Involved in Amiloride Block and Ion Permeation

The amiloride-sensitive epithelial Nachannel (ENaC) is a heteromultimeric channel made of three αβγ subunits. The structures involved in the ion permeation pathway have only been partially identified, and the respective contributions of each subunit in the formation of the conduction pore has not yet been established. Using a site-directed mutagenesis approach, we have identified in a short segment preceding the second membrane-spanning domain (the pre-M2 segment) amino acid residues involved in ion permeation and critical for channel block by amiloride. Cys substitutions of Gly residues in β and γ subunits at position βG525 and γG537 increased the apparent inhibitory constant (K _i) for amiloride by >1,000-fold and decreased channel unitary current without affecting ion selectivity. The corresponding mutation S583 to C in the α subunit increased amiloride K _i by 20-fold, without changing channel conducting properties. Coexpression of these mutated αβγ subunits resulted in a nonconducting channel expressed at the cell surface. Finally, these Cys substitutions increased channel affinity for block by externalZn²⁺ ions, in particular the αS583C mutant showing a K _i for Zn²⁺of 29 μM. Mutations of residues αW582L or βG522D also increased amiloride K _i, the later mutation generating a Ca²⁺blocking site located 15% within the membrane electric field. These experiments provide strong evidence that αβγ ENaCs are pore-forming subunits involved in ion permeation through the channel. The pre-M2 segment of αβγ subunits may form a pore loop structure at the extracellular face of the channel, where amiloride binds within the channel lumen. We propose that amiloride interacts with Na⁺ions at an external Na⁺binding site preventing ion permeation through the channel pore.     

Two Separate Interfaces between the Voltage Sensor and Pore Are Required for the Function of Voltage-Dependent K+ Channels

Voltage-dependent K⁺ (Kv) channels gate open in response to the membrane voltage. To further our understanding of how cell membrane voltage regulates the opening of a Kv channel, we have studied the protein interfaces that attach the voltage-sensor domains to the pore. In the crystal structure, three physical interfaces exist. Only two of these consist of amino acids that are co-evolved across the interface between voltage sensor and pore according to statistical coupling analysis of 360 Kv channel sequences. A first co-evolved interface is formed by the S4-S5 linkers (one from each of four voltage sensors), which form a cuff surrounding the S6-lined pore opening at the intracellular surface. The crystal structure and published mutational studies support the hypothesis that the S4-S5 linkers convert voltage-sensor motions directly into gate opening and closing. A second co-evolved interface forms a small contact surface between S1 of the voltage sensor and the pore helix near the extracellular surface. We demonstrate through mutagenesis that this interface is necessary for the function and/or structure of two different Kv channels. This second interface is well positioned to act as a second anchor point between the voltage sensor and the pore, thus allowing efficient transmission of conformational changes to the pore's gate.     

Kv Channel S1-S2 Linker Working as a Binding Site of Human β-Defensin 2 for Channel Activation Modulation

Among the three extracellular domains of the tetrameric voltage-gated K⁺ (Kv) channels consisting of six membrane-spanning helical segments named S1–S6, the functional role of the S1-S2 linker still remains unclear because of the lack of a peptide ligand. In this study, the Kv1.3 channel S1-S2 linker was reported as a novel receptor site for human β-defensin 2 (hBD2). hBD2 shifts the conductance-voltage relationship curve of the human Kv1.3 channel in a positive direction by nearly 10.5 mV and increases the activation time constant for the channel. Unlike classical gating modifiers of toxin peptides from animal venoms, which generally bind to the Kv channel S3-S4 linker, hBD2 only targets residues in both the N and C termini of the S1-S2 linker to influence channel gating and inhibit channel currents. The increment and decrement of the basic residue number in a positively charged S4 sensor of Kv1.3 channel yields conductance-voltage relationship curves in the positive direction by ∼31.2 mV and 2–4 mV, which suggests that positively charged hBD2 is anchored in the channel S1-S2 linker and is modulating channel activation through electrostatic repulsion with an adjacent S4 helix. Together, these findings reveal a novel peptide ligand that binds with the Kv channel S1-S2 linker to modulate channel activation. These findings also highlight the functional importance of the Kv channel S1-S2 linker in ligand recognition and modification of channel activation.     

Three-dimensional structure of human Kv10.2 ion channel studied by single particle electron microscopy and molecular modeling

Here we present a three-dimensional structure of human voltage gated Kv10.2 ion channel solved at 2.5 nm resolution. We demonstrated that Kv10.2 channel structure is subdivided into two layers. For interpretation of the structure we used the homology modeling, using the transmembrane regions of MlotiK1 channel (C subunit), and cytoplasmic PAS-PAC and cNBD domains of the N-terminal tail of hERG (A subunit) and the bacterial cyclic nucleotide-activated K+ channel binding domain as the templates. The homologous transmembrane part can be fitted into the upper part of the reconstruction. The cytoplasmic domains form the structure, similar to a "hanging gondola", which is connected to the membrane-embedded domain with linkers. The length of linkers allow contacts between C-terminal cNBD domains and N-terminal PAS domains.     

Structural insights into excitation—contraction coupling by electron cryomicroscopy

In muscle, excitation-contraction coupling is defined as the process linking depolarization of the surface membrane with Ca²⁺ release from cytoplasmic stores, which activates contraction of striated muscle. This process is primarily controlled by interplay between two Ca²⁺ channels—the voltage-gated L-type Ca²⁺ channel (dihydropyridine receptor, DHPR) localized in the t-tubule membrane and the Ca²⁺-release channel (ryanodine receptor, RyR) of the sarcoplasmic reticulum membrane. The structures of both channels have been extensively studied by several groups using electron cryomicroscopy and single particle reconstruction techniques. The structures of RyR, determined at resolutions of 22–30 Å, reveal a characteristic mushroom shape with a bulky cytoplasmic region and the membrane-spanning stem. While the cytoplasmic region exhibits a complex structure comprising a multitude of distinctive domains with numerous intervening cavities, at this resolution no definitive statement can be made about the location of the actual pore within the transmembrane region. Conformational changes associated with functional transitions of the Ca²⁺ release channel from closed to open states have been characterized. Further experiments determined localization of binding sites for various channel ligands. The structural studies of the DHPR are less developed. Although four 3D maps of the DHPR were reported recently at 24–30 Å resolution from studies of frozen-hydrated and negatively stained receptors, there are some discrepancies between reported structures with respect to the overall appearance and dimensions of the channel structure. Future structural studies at higher resolution are needed to refine the structures of both channels and to substantiate a proposed molecular model for their interaction.Translated from Biokhimiya, Vol. 69, No. 11, 2004, pp. 1506–1514.Original Russian Text Copyright © 2004 by Serysheva.     

Three-dimensional structure of human voltage-gated ion channel Kv10.2 studied by electron microscopy of macromolecules and molecular modeling

Three-dimensional structure of the human voltage-gated channel Kv10.2 has been elucidated for the first time using the method of electron microscopy with 2.5 nm resolution. The molecule has a distinct domain structure. For interpretation of the structure, homology modeling was used with the cAMP-dependent channel MlotiK1 (C-subunit) structure used as a template for a membrane part of the channel, homology with the structure of the human potassium channel herg (A subunits) was used for the cytoplasmic subdomains PAS-PAC, and for the cNBD domain homology with the MloK1 channel was used. The homologous transmembrane part corresponds by size to the upper part of the three-dimensional reconstruction. Cytoplasmic domains of the Kv10.2 channel form the structure built according to the ‘hanging gondola’ type that is connected with the transmembrane part of the channel by linkers. The length of linkers suggests the possibility of contacts between the C-terminal cNBD domains and N-terminal PAS-domains.     

Molecular movement of the voltage sensor in a K channel

The X-ray crystallographic structure of KvAP, a voltage-gated bacterial K channel, was recently published. However, the position and the molecular movement of the voltage sensor, S4, are still controversial. For example, in the crystallographic structure, S4 is located far away (>30 A) from the pore domain, whereas electrostatic experiments have suggested that S4 is located close (<8 A) to the pore domain in open channels. To test the proposed location and motion of S4 relative to the pore domain, we induced disulphide bonds between pairs of introduced cysteines: one in S4 and one in the pore domain. Several residues in S4 formed a state-dependent disulphide bond with a residue in the pore domain. Our data suggest that S4 is located close to the pore domain in a neighboring subunit. Our data also place constraints on possible models for S4 movement and are not compatible with a recently proposed KvAP model.     

Structural, Biochemical, and Functional Characterization of the Cyclic Nucleotide Binding Homology Domain from the Mouse EAG1 Potassium Channel

KCNH channels are voltage-gated potassium channels with important physiological functions. In these channels, a C-terminal cytoplasmic region, known as the cyclic nucleotide binding homology (CNB-homology) domain displays strong sequence similarity to cyclic nucleotide binding (CNB) domains. However, the isolated domain does not bind cyclic nucleotides. Here, we report the X-ray structure of the CNB-homology domain from the mouse EAG1 channel. Through comparison with the recently determined structure of the CNB-homology domain from the zebrafish ELK (eag-like K(+)) channel and the CNB domains from the MlotiK1 and HCN (hyperpolarization-activated cyclic nucleotide-gated) potassium channels, we establish the structural features of CNB-homology domains that explain the low affinity for cyclic nucleotides. Our structure establishes that the "self-liganded" conformation, where two residues of the C-terminus of the domain are bound in an equivalent position to cyclic nucleotides in CNB domains, is a conserved feature of CNB-homology domains. Importantly, we provide biochemical evidence that suggests that there is also an unliganded conformation where the C-terminus of the domain peels away from its bound position. A functional characterization of this unliganded conformation reveals a role of the CNB-homology domain in channel gating.     

Functional Roles of Clusters of Hydrophobic and Polar Residues in the Epithelial Na+ Channel Knuckle Domain

The extracellular regions of epithelial Na⁺ channel subunits are highly ordered structures composed of domains formed by α helices and β strands. Deletion of the peripheral knuckle domain of the α subunit in the αβγ trimer results in channel activation, reflecting an increase in channel open probability due to a loss of the inhibitory effect of external Na⁺ (Na⁺ self-inhibition). In contrast, deletion of either the β or γ subunit knuckle domain within the αβγ trimer dramatically reduces epithelial Na⁺ channel function and surface expression, and impairs subunit maturation. We systematically mutated individual α subunit knuckle domain residues and assessed functional properties of these mutants. Cysteine substitutions at 14 of 28 residues significantly suppressed Na⁺ self-inhibition. The side chains of a cluster of these residues are non-polar and are predicted to be directed toward the palm domain, whereas a group of polar residues are predicted to orient their side chains toward the space between the knuckle and finger domains. Among the mutants causing the greatest suppression of Na⁺ self-inhibition were αP521C, αI529C, and αS534C. The introduction of Cys residues at homologous sites within either the β or γ subunit knuckle domain resulted in little or no change in Na⁺ self-inhibition. Our results suggest that multiple residues in the α subunit knuckle domain contribute to the mechanism of Na⁺ self-inhibition by interacting with palm and finger domain residues via two separate and chemically distinct motifs.     

Important Role of Asparagines in Coupling the Pore and Votage-Sensor Domain in Voltage-Gated Sodium Channels

Voltage-gated sodium (Na_V) channels contain an α-subunit incorporating the channel’s pore and gating machinery composed of four homologous domains (DI–DIV), with a pore domain formed by the S5 and S6 segments and a voltage-sensor domain formed by the S1–S4 segments. During a membrane depolarization movement, the S4s in the voltage-sensor domains exert downstream effects on the S6 segments to control ionic conductance through the pore domain. We used lidocaine, a local anesthetic and antiarrhythmic drug, to probe the role of conserved Asn residues in the S6s of DIII and DIV in Na_V1.5 and Na_V1.4. Previous studies have shown that lidocaine binding to the pore domain causes a decrease in the maximum gating (Q_max) charge of ∼38%, and three-fourths of this decrease results from the complete stabilization of DIII-S4 (contributing a 30% reduction in Q_max) and one-fourth is due to partial stabilization of DIV-S4 (a reduction of 8–10%). Even though substitutions for the Asn in DIV-S6 in Na_V1.5, N1764A and N1764C, produce little ionic current in transfected mammalian cells, they both express robust gating currents. Anthopleurin-A toxin, which inhibits movement of DIV-S4, still reduced Q_max by nearly 30%, a value similar to that observed in wild-type channels, in both N1764A and N1764C. By applying lidocaine and measuring the gating currents, we demonstrated that Asn residues in the S6s of DIII and DIV are important for coupling their pore domains to their voltage-sensor domains, and that Ala and Cys substitutions for Asn in both S6s result in uncoupling of the pore domains from their voltage-sensor domains. Similar observations were made for Na_V1.4, although substitutions for Asn in DIII-S6 showed somewhat less uncoupling.     

An anion channel of sarcoplasmic reticulum incorporated into planar lipid bilayers: Single-channel behavior and conductance properties

Summary An anion channel of sarcoplasmic reticulum vesicle has been incorporated into planar lipid bilayers by means of a fusion method and its basic properties were investigated. Analysis of fusion processes suggested that one SR vesicle contained approximately one anion channel. The conductance of this channel has several substates and shows a flickering behavior. The occupation probability of each substate was voltage dependent, which induced an inward rectification of macroscopic currents. Further, the anion channel was found to have the following properties. (1) The single-channel conductance is about 200 pS at 100mm Cl^–. (2) The channel does not select among monovalent anions but SO ₄ ^2– hardly permeates through the channel. (3) SO ₄ ^2– added to thecis side (the side to which SR vesicles were added) inhibits Cl^– current competitively in a voltage-dependent manner. (4) An analysis of this voltage dependence suggests that the binding site of SO ₄ ^2– is located at about 36% of the way across the channel from thecis entrance.     

Structural Interactions between Lipids, Water and S1-S4 Voltage-Sensing Domains

Membrane proteins serve crucial signaling and transport functions, yet relatively little is known about their structures in membrane environments or how lipids interact with these proteins. For voltage-activated ion channels, X-ray structures suggest that the mobile voltage-sensing S4 helix would be exposed to the membrane, and functional studies reveal that lipid modification can profoundly alter channel activity. Here, we use solid-state NMR to investigate structural interactions of lipids and water with S1-S4 voltage-sensing domains and to explore whether lipids influence the structure of the protein. Our results demonstrate that S1-S4 domains exhibit extensive interactions with lipids and that these domains are heavily hydrated when embedded in a membrane. We also find evidence for preferential interactions of anionic lipids with S1-S4 domains and that these interactions have lifetimes on the timescale of ≤ 10^− 3 s. Arg residues within S1-S4 domains are well hydrated and are positioned in close proximity to lipids, exhibiting local interactions with both lipid headgroups and acyl chains. Comparative studies with a positively charged lipid lacking a phosphodiester group reveal that this lipid modification has only modest effects on the structure and hydration of S1-S4 domains. Taken together, our results demonstrate that Arg residues in S1-S4 voltage-sensing domains reside in close proximity to the hydrophobic interior of the membrane yet are well hydrated, a requirement for carrying charge and driving protein motions in response to changes in membrane voltage.     

Brownian dynamics investigation into the conductance state of the MscS channel crystal structure

We suggest that the crystal structure of the mechanosensitive channel of small conductance is in a minimally conductive state rather than being fully activated. Performing Brownian dynamics simulations on the crystal structure show that no ions pass through it. When simulations are conducted on just the transmembrane domain (excluding the cytoplasmic residues 128 to 280) ions are seen to pass through the channel, but the conductance of ∼ 30 pS is well below experimentally measured values. The mutation L109S that replaces a pore lining hydrophobic residue with a polar one is found to have little effect on the conductance of the channel. Widening the hydrophobic region of the pore by 2.5 Å however, increases the channel conductance to over 200 pS suggesting that only a minimal conformational change is required to gate the pore.